perpetual 0.9.5__cp310-cp310-win_amd64.whl → 0.10.0__cp310-cp310-win_amd64.whl

perpetual/booster.py

@@ -1,16 +1,19 @@
-import json
 import inspect
+import json
 import warnings
-from typing_extensions import Self
+from types import FunctionType
 from typing import Any, Dict, Iterable, List, Optional, Set, Tuple, Union, cast
 
 import numpy as np
-
-from perpetual.perpetual import PerpetualBooster as CratePerpetualBooster  # type: ignore
-from perpetual.perpetual import MultiOutputBooster as CrateMultiOutputBooster  # type: ignore
+from perpetual.data import Node
+from perpetual.perpetual import (
+    MultiOutputBooster as CrateMultiOutputBooster,  # type: ignore
+)
+from perpetual.perpetual import (
+    PerpetualBooster as CratePerpetualBooster,  # type: ignore
+)
 from perpetual.serialize import BaseSerializer, ObjectSerializer
 from perpetual.types import BoosterType, MultiOutputBoosterType
-from perpetual.data import Node
 from perpetual.utils import (
     CONTRIBUTION_METHODS,
     convert_input_array,
@@ -18,6 +21,7 @@ from perpetual.utils import (
     transform_input_frame,
     type_df,
 )
+from typing_extensions import Self
 
 
 class PerpetualBooster:
@@ -37,7 +41,9 @@ class PerpetualBooster:
     def __init__(
         self,
         *,
-        objective: str = "LogLoss",
+        objective: Union[
+            str, Tuple[FunctionType, FunctionType, FunctionType]
+        ] = "LogLoss",
         budget: float = 0.5,
         num_threads: Optional[int] = None,
         monotone_constraints: Union[Dict[Any, int], None] = None,
@@ -68,6 +74,10 @@ class PerpetualBooster:
                 "QuantileLoss" to use quantile error (regression),
                 "HuberLoss" to use huber error (regression),
                 "AdaptiveHuberLoss" to use adaptive huber error (regression).
+                "ListNetLoss" to use ListNet loss (ranking).
+                custom objective in the form of (grad, hess, init)
+                where grad and hess are functions that take (y, pred, sample_weight, group) and return the gradient and hessian
+                init is a function that takes (y, sample_weight, group) and returns the initial prediction value.
                 Defaults to "LogLoss".
             budget (float, optional): a positive number for fitting budget. Increasing this number will more
                 likely result in more boosting rounds and more increased predictive power.
@@ -165,7 +175,16 @@ class PerpetualBooster:
             {} if monotone_constraints is None else monotone_constraints
         )
 
-        self.objective = objective
+        if isinstance(objective, str):
+            self.objective = objective
+            self.loss = None
+            self.grad = None
+            self.init = None
+        else:
+            self.objective = None
+            self.loss = objective[0]
+            self.grad = objective[1]
+            self.init = objective[2]
         self.budget = budget
         self.num_threads = num_threads
         self.monotone_constraints = monotone_constraints_
@@ -207,10 +226,13 @@ class PerpetualBooster:
             iteration_limit=self.iteration_limit,
             memory_limit=self.memory_limit,
             stopping_rounds=self.stopping_rounds,
+            loss=self.loss,
+            grad=self.grad,
+            init=self.init,
         )
         self.booster = cast(BoosterType, booster)
 
-    def fit(self, X, y, sample_weight=None) -> Self:
+    def fit(self, X, y, sample_weight=None, group=None) -> Self:
         """Fit the gradient booster on a provided dataset.
 
         Args:
@@ -220,11 +242,19 @@ class PerpetualBooster:
             sample_weight (Union[ArrayLike, None], optional): Instance weights to use when
                 training the model. If None is passed, a weight of 1 will be used for every record.
                 Defaults to None.
+            group (Union[ArrayLike, None], optional): Group lengths to use for a ranking objective.
+                If None is passes, all items are assumed to be in the same group.
+                Defaults to None.
         """
 
-        features_, flat_data, rows, cols, categorical_features_, cat_mapping = (
-            convert_input_frame(X, self.categorical_features, self.max_cat)
-        )
+        (
+            features_,
+            flat_data,
+            rows,
+            cols,
+            categorical_features_,
+            cat_mapping,
+        ) = convert_input_frame(X, self.categorical_features, self.max_cat)
         self.n_features_ = cols
         self.cat_mapping = cat_mapping
         self.feature_names_in_ = features_
@@ -237,6 +267,11 @@ class PerpetualBooster:
         else:
             sample_weight_, _ = convert_input_array(sample_weight, self.objective)
 
+        if group is None:
+            group_ = None
+        else:
+            group_, _ = convert_input_array(group, self.objective, is_int=True)
+
         # Convert the monotone constraints into the form needed
         # by the rust code.
         crate_mc = self._standardize_monotonicity_map(X)
@@ -265,6 +300,9 @@ class PerpetualBooster:
                 iteration_limit=self.iteration_limit,
                 memory_limit=self.memory_limit,
                 stopping_rounds=self.stopping_rounds,
+                loss=self.loss,
+                grad=self.grad,
+                init=self.init,
             )
             self.booster = cast(BoosterType, booster)
         else:
@@ -289,6 +327,9 @@ class PerpetualBooster:
                 iteration_limit=self.iteration_limit,
                 memory_limit=self.memory_limit,
                 stopping_rounds=self.stopping_rounds,
+                loss=self.loss,
+                grad=self.grad,
+                init=self.init,
             )
             self.booster = cast(MultiOutputBoosterType, booster)
 
@@ -308,11 +349,12 @@ class PerpetualBooster:
             cols=cols,
             y=y_,
             sample_weight=sample_weight_,  # type: ignore
+            group=group_,
         )
 
         return self
 
-    def prune(self, X, y, sample_weight=None) -> Self:
+    def prune(self, X, y, sample_weight=None, group=None) -> Self:
         """Prune the gradient booster on a provided dataset.
 
         Args:
@@ -322,6 +364,9 @@ class PerpetualBooster:
             sample_weight (Union[ArrayLike, None], optional): Instance weights to use when
                 training the model. If None is passed, a weight of 1 will be used for every record.
                 Defaults to None.
+            group (Union[ArrayLike, None], optional): Group lengths to use for a ranking objective.
+                If None is passes, all items are assumed to be in the same group.
+                Defaults to None.
         """
 
         _, flat_data, rows, cols = transform_input_frame(X, self.cat_mapping)
@@ -333,18 +378,24 @@ class PerpetualBooster:
         else:
             sample_weight_, _ = convert_input_array(sample_weight, self.objective)
 
+        if group is None:
+            group_ = None
+        else:
+            group_, _ = convert_input_array(group, self.objective, is_int=True)
+
         self.booster.prune(
             flat_data=flat_data,
             rows=rows,
             cols=cols,
             y=y_,
             sample_weight=sample_weight_,  # type: ignore
+            group=group_,
         )
 
         return self
 
     def calibrate(
-        self, X_train, y_train, X_cal, y_cal, alpha, sample_weight=None
+        self, X_train, y_train, X_cal, y_cal, alpha, sample_weight=None, group=None
     ) -> Self:
         """Calibrate the gradient booster on a provided dataset.
 
@@ -361,6 +412,9 @@ class PerpetualBooster:
             sample_weight (Union[ArrayLike, None], optional): Instance weights to use when
                 training the model. If None is passed, a weight of 1 will be used for every record.
                 Defaults to None.
+            group (Union[ArrayLike, None], optional): Group lengths to use for a ranking objective.
+                If None is passes, all items are assumed to be in the same group.
+                Defaults to None.
         """
 
         _, flat_data_train, rows_train, cols_train = transform_input_frame(
@@ -391,6 +445,7 @@ class PerpetualBooster:
             y_cal=y_cal_,
             alpha=np.array(alpha),
             sample_weight=sample_weight_,  # type: ignore
+            group=group,
         )
 
         return self

perpetual/sklearn.py

@@ -0,0 +1,193 @@
+import warnings
+from types import FunctionType
+from typing import Any, Dict, Optional, Tuple, Union
+
+from perpetual.booster import PerpetualBooster
+from sklearn.base import ClassifierMixin, RegressorMixin
+from sklearn.metrics import accuracy_score, r2_score
+from typing_extensions import Self
+
+
+class PerpetualClassifier(PerpetualBooster, ClassifierMixin):
+    """
+    A scikit-learn compatible classifier based on PerpetualBooster.
+    Uses 'LogLoss' as the default objective.
+    """
+
+    # Expose the objective explicitly in the __init__ signature to allow
+    # scikit-learn to correctly discover and set it via set_params.
+    def __init__(
+        self,
+        *,
+        objective: Union[
+            str, Tuple[FunctionType, FunctionType, FunctionType]
+        ] = "LogLoss",
+        budget: float = 0.5,
+        num_threads: Optional[int] = None,
+        monotone_constraints: Union[Dict[Any, int], None] = None,
+        # ... other parameters ...
+        max_bin: int = 256,
+        max_cat: int = 1000,
+        # Capture all parameters in a way that BaseEstimator can handle
+        **kwargs,
+    ):
+        # Ensure the objective is one of the valid classification objectives
+        valid_objectives = {
+            "LogLoss"
+        }  # Assuming only LogLoss for classification for simplicity
+        if isinstance(objective, str) and objective not in valid_objectives:
+            # Custom objectives are allowed via the tuple form
+            pass
+
+        super().__init__(
+            objective=objective,
+            budget=budget,
+            num_threads=num_threads,
+            monotone_constraints=monotone_constraints,
+            # ... pass all other parameters ...
+            max_bin=max_bin,
+            max_cat=max_cat,
+            **kwargs,  # Catch-all for any other parameters passed by user or set_params
+        )
+
+    # fit, predict, predict_proba, and predict_log_proba are inherited
+    # and properly adapted in PerpetualBooster.
+
+    def score(self, X, y, sample_weight=None):
+        """Returns the mean accuracy on the given test data and labels."""
+        preds = self.predict(X)
+        return accuracy_score(y, preds, sample_weight=sample_weight)
+
+    def fit(self, X, y, sample_weight=None, **fit_params) -> Self:
+        """A wrapper for the base fit method."""
+        # Check if objective is appropriate for classification if it's a string
+        if isinstance(self.objective, str) and self.objective not in ["LogLoss"]:
+            warnings.warn(
+                f"Objective '{self.objective}' is typically for regression/ranking but used in PerpetualClassifier. Consider 'LogLoss'."
+            )
+
+        # In classification, the labels (classes_) are set in the base fit.
+        return super().fit(X, y, sample_weight=sample_weight, **fit_params)
+
+
+class PerpetualRegressor(PerpetualBooster, RegressorMixin):
+    """
+    A scikit-learn compatible regressor based on PerpetualBooster.
+    Uses 'SquaredLoss' as the default objective.
+    """
+
+    def __init__(
+        self,
+        *,
+        objective: Union[
+            str, Tuple[FunctionType, FunctionType, FunctionType]
+        ] = "SquaredLoss",
+        budget: float = 0.5,
+        num_threads: Optional[int] = None,
+        monotone_constraints: Union[Dict[Any, int], None] = None,
+        # ... other parameters ...
+        max_bin: int = 256,
+        max_cat: int = 1000,
+        **kwargs,
+    ):
+        # Enforce or warn about regression objectives
+        valid_objectives = {
+            "SquaredLoss",
+            "QuantileLoss",
+            "HuberLoss",
+            "AdaptiveHuberLoss",
+        }
+        if isinstance(objective, str) and objective not in valid_objectives:
+            pass  # Allow for custom string or tuple objective
+
+        super().__init__(
+            objective=objective,
+            budget=budget,
+            num_threads=num_threads,
+            monotone_constraints=monotone_constraints,
+            # ... pass all other parameters ...
+            max_bin=max_bin,
+            max_cat=max_cat,
+            **kwargs,
+        )
+
+    def fit(self, X, y, sample_weight=None, **fit_params) -> Self:
+        """A wrapper for the base fit method."""
+        # For regression, we typically enforce len(self.classes_) == 0 after fit
+        if isinstance(self.objective, str) and self.objective not in [
+            "SquaredLoss",
+            "QuantileLoss",
+            "HuberLoss",
+            "AdaptiveHuberLoss",
+        ]:
+            warnings.warn(
+                f"Objective '{self.objective}' may not be suitable for PerpetualRegressor. Consider 'SquaredLoss' or a quantile/huber loss."
+            )
+
+        return super().fit(X, y, sample_weight=sample_weight, **fit_params)
+
+    def score(self, X, y, sample_weight=None):
+        """Returns the coefficient of determination ($R^2$) of the prediction."""
+        preds = self.predict(X)
+        return r2_score(y, preds, sample_weight=sample_weight)
+
+
+class PerpetualRanker(
+    PerpetualBooster, RegressorMixin
+):  # Ranking models sometimes inherit from RegressorMixin for compatibility
+    """
+    A scikit-learn compatible ranker based on PerpetualBooster.
+    Uses 'ListNetLoss' as the default objective.
+    Requires the 'group' parameter to be passed to fit.
+    """
+
+    def __init__(
+        self,
+        *,
+        objective: Union[
+            str, Tuple[FunctionType, FunctionType, FunctionType]
+        ] = "ListNetLoss",
+        budget: float = 0.5,
+        num_threads: Optional[int] = None,
+        monotone_constraints: Union[Dict[Any, int], None] = None,
+        # ... other parameters ...
+        max_bin: int = 256,
+        max_cat: int = 1000,
+        **kwargs,
+    ):
+        if isinstance(objective, str) and objective not in {"ListNetLoss"}:
+            warnings.warn(
+                f"Objective '{objective}' may not be suitable for PerpetualRanker. Consider 'ListNetLoss'."
+            )
+
+        super().__init__(
+            objective=objective,
+            budget=budget,
+            num_threads=num_threads,
+            monotone_constraints=monotone_constraints,
+            # ... pass all other parameters ...
+            max_bin=max_bin,
+            max_cat=max_cat,
+            **kwargs,
+        )
+
+    def fit(self, X, y, group=None, sample_weight=None, **fit_params) -> Self:
+        """
+        Fit the ranker. Requires the 'group' parameter.
+
+        Args:
+            X: Training data.
+            y: Target relevance scores.
+            group: Group lengths to use for a ranking objective. (Required for ListNetLoss).
+            sample_weight: Instance weights.
+        """
+        if (
+            group is None
+            and isinstance(self.objective, str)
+            and self.objective == "ListNetLoss"
+        ):
+            raise ValueError(
+                "The 'group' parameter must be provided when using the 'ListNetLoss' objective for ranking."
+            )
+
+        return super().fit(X, y, sample_weight=sample_weight, group=group, **fit_params)

perpetual/utils.py

@@ -1,7 +1,7 @@
 import logging
-import numpy as np
 from typing import Dict, Iterable, List, Optional, Tuple
 
+import numpy as np
 
 logger = logging.getLogger(__name__)
 
@@ -32,7 +32,7 @@ def type_series(y):
         return ""
 
 
-def convert_input_array(x, objective, is_target=False) -> np.ndarray:
+def convert_input_array(x, objective, is_target=False, is_int=False) -> np.ndarray:
     classes_ = []
 
     if type(x).__module__.split(".")[0] == "numpy":
@@ -55,7 +55,10 @@ def convert_input_array(x, objective, is_target=False) -> np.ndarray:
         if len(classes_) > 2:
             x_ = np.squeeze(np.eye(len(classes_))[x_index])
 
-    if not np.issubdtype(x_.dtype, "float64"):
+    if is_int and not np.issubdtype(x_.dtype, "uint64"):
+        x_ = x_.astype(dtype="uint64", copy=False)
+
+    if not is_int and not np.issubdtype(x_.dtype, "float64"):
         x_ = x_.astype(dtype="float64", copy=False)
 
     if len(x_.shape) == 2:

perpetual-0.10.0.dist-info/METADATA

@@ -0,0 +1,31 @@
+Metadata-Version: 2.4
+Name: perpetual
+Version: 0.10.0
+Classifier: Programming Language :: Rust
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Requires-Dist: numpy
+Requires-Dist: typing-extensions
+Requires-Dist: black ; extra == 'dev'
+Requires-Dist: pandas ; extra == 'dev'
+Requires-Dist: polars ; extra == 'dev'
+Requires-Dist: pyarrow ; extra == 'dev'
+Requires-Dist: maturin ; extra == 'dev'
+Requires-Dist: pytest ; extra == 'dev'
+Requires-Dist: seaborn ; extra == 'dev'
+Requires-Dist: scikit-learn ; extra == 'dev'
+Requires-Dist: mkdocs-material ; extra == 'dev'
+Requires-Dist: mkdocstrings[python] ; extra == 'dev'
+Requires-Dist: mkdocs-autorefs ; extra == 'dev'
+Requires-Dist: ruff ; extra == 'dev'
+Provides-Extra: dev
+License-File: LICENSE
+Summary: A self-generalizing gradient boosting machine that doesn't need hyperparameter optimization
+Keywords: rust,perpetual,machine learning,tree model,decision tree,gradient boosted decision tree,gradient boosting machine
+Home-Page: https://perpetual-ml.com
+Author-email: Mutlu Simsek <mutlusims3k@gmail.com>, Serkan Korkmaz <serkor1@duck.com>, Pieter Pel <pelpieter@gmail.com>
+Requires-Python: >=3.9

perpetual-0.10.0.dist-info/RECORD

@@ -0,0 +1,12 @@
+perpetual-0.10.0.dist-info/METADATA,sha256=jz1ubQqMaGY-CjOIcUJdfzP2sTlikFLMShWkEXlG10s,1403
+perpetual-0.10.0.dist-info/WHEEL,sha256=Iz7QqxpWQRXToFIDkGspPPKDuV_klwuhW8ziiU5jhR8,96
+perpetual-0.10.0.dist-info/licenses/LICENSE,sha256=gcuuhKKc5-dwvyvHsXjlC9oM6N5gZ6umYbC8ewW1Yvg,35821
+perpetual/__init__.py,sha256=V0RhghaG0CuKxKrzYUBYqrf7Drb-gjmznsbz9KT12lk,122
+perpetual/booster.py,sha256=ZPymfG5L1M8XTld1H4af6k61T3eHPrbToTVFDHH29Ro,53161
+perpetual/data.py,sha256=vhjWEc_ESYWoaczz0GkUPtfS0iRSKdVZSrCkQn8yLPw,630
+perpetual/perpetual.cp310-win_amd64.pyd,sha256=FPo7v23dIgR8sPAW5cLnfuYu4gvpUgARsphV8D_A2g8,1767936
+perpetual/serialize.py,sha256=FeW4JsUFVsrft9N7gz-ebn5mXvDv4LiJC2sgBEeGxYo,1957
+perpetual/sklearn.py,sha256=5d1clRslX4-Kt8DwE-Jht9xZ01VeSNnz_ZmXWvkg8lc,7203
+perpetual/types.py,sha256=idZNsDErNTur_rJ_5Co8Pb6fik-AUn9lkrXmjbQJVX0,3381
+perpetual/utils.py,sha256=IiQtM6v7Ve4GNuKhjiAHuGal0QPoYG7CI55q_Ci3yd4,7627
+perpetual-0.10.0.dist-info/RECORD,,

{perpetual-0.9.5.dist-info → perpetual-0.10.0.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: maturin (1.9.1)
+Generator: maturin (1.9.4)
 Root-Is-Purelib: false
 Tag: cp310-cp310-win_amd64

perpetual-0.9.5.dist-info/METADATA

@@ -1,166 +0,0 @@
-Metadata-Version: 2.4
-Name: perpetual
-Version: 0.9.5
-Classifier: Programming Language :: Rust
-Classifier: Programming Language :: Python :: 3
-Classifier: Programming Language :: Python :: 3.9
-Classifier: Programming Language :: Python :: 3.10
-Classifier: Programming Language :: Python :: 3.11
-Classifier: Programming Language :: Python :: 3.12
-Classifier: Programming Language :: Python :: 3.13
-Requires-Dist: numpy
-Requires-Dist: typing-extensions
-Requires-Dist: black ; extra == 'dev'
-Requires-Dist: pandas ; extra == 'dev'
-Requires-Dist: polars ; extra == 'dev'
-Requires-Dist: pyarrow ; extra == 'dev'
-Requires-Dist: maturin ; extra == 'dev'
-Requires-Dist: pytest ; extra == 'dev'
-Requires-Dist: seaborn ; extra == 'dev'
-Requires-Dist: scikit-learn ; extra == 'dev'
-Requires-Dist: mkdocs-material ; extra == 'dev'
-Requires-Dist: mkdocstrings[python] ; extra == 'dev'
-Requires-Dist: mkdocs-autorefs ; extra == 'dev'
-Requires-Dist: ruff ; extra == 'dev'
-Provides-Extra: dev
-License-File: LICENSE
-Summary: A self-generalizing gradient boosting machine that doesn't need hyperparameter optimization
-Keywords: rust,perpetual,machine learning,tree model,decision tree,gradient boosted decision tree,gradient boosting machine
-Home-Page: https://perpetual-ml.com
-Author: Mutlu Simsek
-Author-email: Mutlu Simsek <msimsek@perpetual-ml.com>
-Requires-Python: >=3.9
-Description-Content-Type: text/markdown; charset=UTF-8; variant=GFM
-Project-URL: Source Code, https://github.com/perpetual-ml/perpetual
-
-<p align="center">
-  <img  height="120" src="https://github.com/perpetual-ml/perpetual/raw/main/resources/perp_logo.png">
-</p>
-
-<div align="center">
-
-[![Python Versions](https://img.shields.io/pypi/pyversions/perpetual.svg?logo=python&logoColor=white)](https://pypi.org/project/perpetual)
-[![PyPI Version](https://img.shields.io/pypi/v/perpetual.svg?logo=pypi&logoColor=white)](https://pypi.org/project/perpetual)
-[![Crates.io Version](https://img.shields.io/crates/v/perpetual?logo=rust&logoColor=white)](https://crates.io/crates/perpetual)
-[![Static Badge](https://img.shields.io/badge/join-discord-blue?logo=discord)](https://discord.gg/AyUK7rr6wy)
-![PyPI - Downloads](https://img.shields.io/pypi/dm/perpetual)
-
-
-</div>
-
-# Perpetual
-
-PerpetualBooster is a gradient boosting machine (GBM) algorithm that doesn't need hyperparameter optimization unlike other GBM algorithms. Similar to AutoML libraries, it has a `budget` parameter. Increasing the `budget` parameter increases the predictive power of the algorithm and gives better results on unseen data. Start with a small budget (e.g. 0.5) and increase it (e.g. 1.0) once you are confident with your features. If you don't see any improvement with further increasing the `budget`, it means that you are already extracting the most predictive power out of your data.
-
-## Usage
-
-You can use the algorithm like in the example below. Check examples folders for both Rust and Python.
-
-```python
-from perpetual import PerpetualBooster
-
-model = PerpetualBooster(objective="SquaredLoss", budget=0.5)
-model.fit(X, y)
-```
-
-## Documentation
-
-Documentation for the Python API can be found [here](https://perpetual-ml.github.io/perpetual) and for the Rust API [here](https://docs.rs/perpetual/latest/perpetual/).
-
-
-## Benchmark
-
-### PerpetualBooster vs. Optuna + LightGBM
-
-Hyperparameter optimization usually takes 100 iterations with plain GBM algorithms. PerpetualBooster achieves the same accuracy in a single run. Thus, it achieves up to 100x speed-up at the same accuracy with different `budget` levels and with different datasets.
-
-The following table summarizes the results for the [California Housing](https://scikit-learn.org/stable/modules/generated/sklearn.datasets.fetch_california_housing.html) dataset (regression):
-
-| Perpetual budget | LightGBM n_estimators | Perpetual mse | LightGBM mse | Speed-up wall time | Speed-up cpu time |
-| ---------------- | --------------------- | ------------- | ------------ | ------------------ | ----------------- |
-| 1.0              | 100                   | 0.192         | 0.192        | 54x                | 56x               |
-| 1.5              | 300                   | 0.188         | 0.188        | 59x                | 58x               |
-| 2.1              | 1000                  | 0.185         | 0.186        | 42x                | 41x               |
-
-The following table summarizes the results for the [Cover Types](https://scikit-learn.org/stable/modules/generated/sklearn.datasets.fetch_covtype.html) dataset (classification):
-
-| Perpetual budget | LightGBM n_estimators | Perpetual log loss | LightGBM log loss | Speed-up wall time | Speed-up cpu time |
-| ---------------- | --------------------- | ------------------ | ----------------- | ------------------ | ----------------- |
-| 0.9              | 100                   | 0.091              | 0.084             | 72x                | 78x               |
-
-The results can be reproduced using the scripts in the [examples](./python-package/examples) folder.
-
-### PerpetualBooster vs. AutoGluon
-
-PerpetualBooster is a GBM but behaves like AutoML so it is benchmarked also against AutoGluon (v1.2, best quality preset), the current leader in [AutoML benchmark](https://automlbenchmark.streamlit.app/cd_diagram). Top 10 datasets with the most number of rows are selected from [OpenML datasets](https://www.openml.org/) for both regression and classification tasks. 
-
-The results are summarized in the following table for regression tasks:
-
-| OpenML Task | Perpetual Training Duration | Perpetual Inference Duration | Perpetual RMSE | AutoGluon Training Duration | AutoGluon Inference Duration | AutoGluon RMSE |
-| -------------------------------------------------------- | ----- | ----- | ------------------- | -------- | ------ | ------------------ |
-| [Airlines_DepDelay_10M](https://www.openml.org/t/359929) | 518   | 11.3  | 29.0                | 520      | 30.9   | <ins> 28.8 </ins>  |
-| [bates_regr_100](https://www.openml.org/t/361940)        | 3421  | 15.1  | <ins> 1.084 </ins>  | OOM      | OOM    | OOM                |
-| [BNG(libras_move)](https://www.openml.org/t/7327)        | 1956  | 4.2   | <ins> 2.51 </ins>   | 1922     | 97.6   | 2.53               |
-| [BNG(satellite_image)](https://www.openml.org/t/7326)    | 334   | 1.6   | 0.731               | 337      | 10.0   | <ins> 0.721 </ins> |
-| [COMET_MC](https://www.openml.org/t/14949)               | 44    | 1.0   | <ins> 0.0615 </ins> | 47       | 5.0    | 0.0662             |
-| [friedman1](https://www.openml.org/t/361939)             | 275   | 4.2   | <ins> 1.047 </ins>  | 278      | 5.1    | 1.487              |
-| [poker](https://www.openml.org/t/10102)                  | 38    | 0.6   | <ins> 0.256 </ins>  | 41       | 1.2    | 0.722              |
-| [subset_higgs](https://www.openml.org/t/361955)          | 868   | 10.6  | <ins> 0.420 </ins>  | 870      | 24.5   | 0.421              |
-| [BNG(autoHorse)](https://www.openml.org/t/7319)          | 107   | 1.1   | <ins> 19.0 </ins>   | 107      | 3.2    | 20.5               |
-| [BNG(pbc)](https://www.openml.org/t/7318)                | 48    | 0.6   | <ins> 836.5 </ins>  | 51       | 0.2    | 957.1              |
-| average                                                  | 465   | 3.9   | -                   | 464      | 19.7   | -                  |
-
-PerpetualBooster outperformed AutoGluon on 8 out of 10 regression tasks, training equally fast and inferring 5.1x faster. 
-
-The results are summarized in the following table for classification tasks:
-
-| OpenML Task | Perpetual Training Duration | Perpetual Inference Duration | Perpetual AUC | AutoGluon Training Duration | AutoGluon Inference Duration | AutoGluon AUC |
-| -------------------------------------------------------- | ------- | ------ | ------------------- | -------- | ------ | ------------------ |
-| [BNG(spambase)](https://www.openml.org/t/146163)         | 70.1    | 2.1   | <ins> 0.671 </ins> | 73.1     | 3.7    | 0.669              |
-| [BNG(trains)](https://www.openml.org/t/208)              | 89.5    | 1.7   | <ins> 0.996 </ins> | 106.4    | 2.4    | 0.994              |
-| [breast](https://www.openml.org/t/361942)                | 13699.3 | 97.7  | <ins> 0.991 </ins> | 13330.7  | 79.7   | 0.949              |
-| [Click_prediction_small](https://www.openml.org/t/7291)  | 89.1    | 1.0   | <ins> 0.749 </ins> | 101.0    | 2.8    | 0.703              |
-| [colon](https://www.openml.org/t/361938)                 | 12435.2 | 126.7 | <ins> 0.997 </ins> | 12356.2  | 152.3  | 0.997              |
-| [Higgs](https://www.openml.org/t/362113)                 | 3485.3  | 40.9  | <ins> 0.843 </ins> | 3501.4   | 67.9   | 0.816              |
-| [SEA(50000)](https://www.openml.org/t/230)               | 21.9    | 0.2   | <ins> 0.936 </ins> | 25.6     | 0.5    | 0.935              |
-| [sf-police-incidents](https://www.openml.org/t/359994)   | 85.8    | 1.5   | <ins> 0.687 </ins> | 99.4     | 2.8    | 0.659              |
-| [bates_classif_100](https://www.openml.org/t/361941)     | 11152.8 | 50.0  | <ins> 0.864 </ins> | OOM      | OOM    | OOM                |
-| [prostate](https://www.openml.org/t/361945)              | 13699.9 | 79.8  | <ins> 0.987 </ins> | OOM      | OOM    | OOM                |
-| average                                                  | 3747.0  | 34.0  | -                  | 3699.2   | 39.0   | -                  |
-
-PerpetualBooster outperformed AutoGluon on 10 out of 10 classification tasks, training equally fast and inferring 1.1x faster. 
-
-PerpetualBooster demonstrates greater robustness compared to AutoGluon, successfully training on all 20 tasks, whereas AutoGluon encountered out-of-memory errors on 3 of those tasks.
-
-The results can be reproduced using the automlbenchmark fork [here](https://github.com/deadsoul44/automlbenchmark).
-
-
-
-## Installation
-
-The package can be installed directly from [pypi](https://pypi.org/project/perpetual):
-
-```shell
-pip install perpetual
-```
-
-Using [conda-forge](https://anaconda.org/conda-forge/perpetual):
-
-```shell
-conda install conda-forge::perpetual
-```
-
-To use in a Rust project and to get the package from [crates.io](https://crates.io/crates/perpetual):
-
-```shell
-cargo add perpetual
-```
-
-## Contribution
-
-Contributions are welcome. Check CONTRIBUTING.md for the guideline.
-
-## Paper
-
-PerpetualBooster prevents overfitting with a generalization algorithm. The paper is work-in-progress to explain how the algorithm works. Check our [blog post](https://perpetual-ml.com/blog/how-perpetual-works) for a high level introduction to the algorithm.
-

perpetual-0.9.5.dist-info/RECORD

@@ -1,11 +0,0 @@
-perpetual-0.9.5.dist-info/METADATA,sha256=Sn-DvsBa-8pKP2NRhJMlVvh_fq0aeF19nI7NgsW7C6s,10724
-perpetual-0.9.5.dist-info/WHEEL,sha256=QKV4Sl7MSpV78xozy9-tp6UITYpirrRleZnt0vKJJXI,96
-perpetual-0.9.5.dist-info/licenses/LICENSE,sha256=gcuuhKKc5-dwvyvHsXjlC9oM6N5gZ6umYbC8ewW1Yvg,35821
-perpetual/__init__.py,sha256=V0RhghaG0CuKxKrzYUBYqrf7Drb-gjmznsbz9KT12lk,122
-perpetual/booster.py,sha256=vyZxchCqvPV79At-yoOVMLvCGdv8xISk2wq_Yu90DrI,50929
-perpetual/data.py,sha256=vhjWEc_ESYWoaczz0GkUPtfS0iRSKdVZSrCkQn8yLPw,630
-perpetual/perpetual.cp310-win_amd64.pyd,sha256=djVhR2-4tl_7IPmbIr8xYkYVo5mNXFI1jzBAHtLHoWo,1723392
-perpetual/serialize.py,sha256=FeW4JsUFVsrft9N7gz-ebn5mXvDv4LiJC2sgBEeGxYo,1957
-perpetual/types.py,sha256=idZNsDErNTur_rJ_5Co8Pb6fik-AUn9lkrXmjbQJVX0,3381
-perpetual/utils.py,sha256=2ifo-9OXaeZBevSo0HKN4uKVy5qT4LqRAchrtZa9yMM,7486
-perpetual-0.9.5.dist-info/RECORD,,